SQLiteAsyncConnection does not release database file

[gentledepp]

Hi!

I made a PR to support storing guids as blobs #739

When running the tests, I noticed, that the async tests fail - except of the very first one.
The exception that causes the UTs to fail is

SetUp : System.IO.IOException : The process cannot access the file '......async.db' because it is being used by another process

Note, that I am running these tests with
USE_SQLITEPCL_RAW

it seems that everything is disposed correctly in the tests.
Then I found [this issue on stackoverflow] (https://stackoverflow.com/questions/8511901/system-data-sqlite-close-not-releasing-database-file) which sounded pretty similar.
Explanation:

What happens when you call SQLiteConnection.Close() is that (along with a number of checks and other things) the SQLiteConnectionHandle that points to the SQLite database instance is disposed. This is done through a call to SQLiteConnectionHandle.Dispose(), however this doesn't actually release the pointer until the CLR's Garbage Collector performs some garbage collection. Since SQLiteConnectionHandle overrides the CriticalHandle.ReleaseHandle() function to call sqlite3_close_interop() (through another function) this does not close the database.

So I tried out their solution, which is adding a call to "GC.Collect" in order to be able to delete the file.

        [SetUp]
    	public void SetUp()
    	{
    		SQLite.SQLiteConnectionPool.Shared.Reset ();
        #if NETFX_CORE
        //...
        #else
    		_connectionString = Path.Combine (Path.GetTempPath (), DatabaseName);
    		_path = _connectionString;
    		GC.Collect ();
    		GC.WaitForPendingFinalizers ();
    		System.IO.File.Delete (_path);
        #endif
    	}

This actually did the trick, however I am not sure in how to proceed...
@ericsink and @praeclarum:

1. is this a known issue of SQLite.PCL.raw?
2. Is it intended? If so, how do I fix it
3. I tried to upgrade to SQLitePCLRaw v1.1.11 but that did not change anything

[praeclarum]

Sorry for the trouble. I haven't thought too hard around the semantics of Resetting the pool, but what you're trying to do look valid to me and I'd like to support it.

Will have to dive into @ericsink's code to find out why the GC needs to run...

[Weldryn]

I happen to run into this issue too. After resetting the pool, setting the connection instance to null and even forcing the GC to run like in @gentledepp example, I can't create a file stream (open mode, read only and read/write access) to the database (for copy/move). Even after 5 attempts with 200ms delay in between. I am running the version 1.3.2 though.
Any news on that bug @praeclarum? Is that fixed in the newest version?

[praeclarum]

I believe the cause was the read connection opened by the async connection. I have changed the code to fix it (tests pass). I will do a release today and ask that you test against that.

Thanks!

[praeclarum]

This has been fixed in 1.6.

[gentledepp]

@bruzkovsky - we need to have a look at this!
@praeclarum - thank you so much for your awesome work.